The DicomDecompression
stage of the latest CTP version will erroneously leave an 8-byte sequence
delimiter in a DICOM file, making the file unreadable.
Symptom:
Processed DICOM cannot
be opened by many dicom readers due to invalid formatting. For instance,
Offis dcmdump yields:
E: DcmItem: Parse error
in sequence item, found (fffe,e0dd) instead of item delimiter (fffe,e00d)
E: dcmdump: Item Delimitation Item missing: reading file: processed_dicom
Cause:
The direct cause of the
symptom is that the last 8 bytes in the file should not be there. They encode
the DICOM sequence termination item fffe,e0dd. The
exact byte sequence in hex is ff fe e0 dd 00 00 00 00. This byte sequence should have been removed but it has not.
The underlying issue is somewhere in the DicomDecompression step of CTP. The original data contains a sequence pixeldata element:
The bug is NOT present
in this CTP version:
2022.03.26 at 11:21:47
CD
All our systems are currently downgraded to the older CTP version to be able to process our data. Is this bug something that can be fixed in java CTP?